CN115147417B - Function connection matrix processing system and device based on filtering method feature selection - Google Patents
Function connection matrix processing system and device based on filtering method feature selection Download PDFInfo
- Publication number
- CN115147417B CN115147417B CN202211070002.9A CN202211070002A CN115147417B CN 115147417 B CN115147417 B CN 115147417B CN 202211070002 A CN202211070002 A CN 202211070002A CN 115147417 B CN115147417 B CN 115147417B
- Authority
- CN
- China
- Prior art keywords
- matrix
- brain
- module
- sel
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000011159 matrix material Substances 0.000 title claims abstract description 99
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000012545 processing Methods 0.000 title claims abstract description 39
- 238000001914 filtration Methods 0.000 title claims abstract description 29
- 230000006870 function Effects 0.000 claims abstract description 42
- 201000010099 disease Diseases 0.000 claims abstract description 37
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 37
- 238000003745 diagnosis Methods 0.000 claims abstract description 34
- 230000003925 brain function Effects 0.000 claims abstract description 18
- 230000000284 resting effect Effects 0.000 claims abstract description 15
- 210000004556 brain Anatomy 0.000 claims description 53
- 238000012360 testing method Methods 0.000 claims description 30
- 238000007781 pre-processing Methods 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 17
- 238000000605 extraction Methods 0.000 claims description 9
- 238000010801 machine learning Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 238000013136 deep learning model Methods 0.000 claims description 5
- 238000012937 correction Methods 0.000 claims description 4
- 238000009499 grossing Methods 0.000 claims description 4
- 230000033001 locomotion Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000002123 temporal effect Effects 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims 1
- 208000020016 psychiatric disease Diseases 0.000 description 7
- 230000000694 effects Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 238000000540 analysis of variance Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000002599 functional magnetic resonance imaging Methods 0.000 description 3
- 238000002610 neuroimaging Methods 0.000 description 3
- 230000001149 cognitive effect Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000002829 reductive effect Effects 0.000 description 2
- 210000003625 skull Anatomy 0.000 description 2
- 206010003805 Autism Diseases 0.000 description 1
- 208000020706 Autistic disease Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000000546 chi-square test Methods 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000004886 head movement Effects 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000002595 magnetic resonance imaging Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 238000009126 molecular therapy Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008506 pathogenesis Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000008733 trauma Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
- G06T7/0014—Biomedical image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
- G06T7/0014—Biomedical image inspection using an image reference approach
- G06T7/0016—Biomedical image inspection using an image reference approach involving temporal comparison
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
- A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/771—Feature selection, e.g. selecting representative features from a multi-dimensional feature space
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Radiology & Medical Imaging (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Quality & Reliability (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Mathematical Optimization (AREA)
- Veterinary Medicine (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Public Health (AREA)
- Multimedia (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- High Energy & Nuclear Physics (AREA)
- Algebra (AREA)
- Fuzzy Systems (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
Abstract
The invention discloses a function connection matrix processing system and device based on filtering method feature selection, comprising: obtaining a resting state brain function magnetic resonance image after being pretreated; extracting a time sequence; calculating a Pearson correlation coefficient to obtain a Pearson correlation coefficient matrix; vectorizing the Pearson correlation coefficient matrix; calculating quantitative correlation indexes by using a filtering method, and selecting the quantitative correlation indexes based on a preset threshold value; weighting the selected functional connection characteristics by using the corresponding quantitative correlation indexes with high correlation with the disease diagnosis result to obtain a functional connection matrix; and obtaining a prediction result through the function connection matrix. On the basis of selecting the characteristics by a filtering method, the selected characteristics are weighted by the quantitative correlation indexes of the calculated characteristics and the disease diagnosis results, so that the characteristics with high correlation with the disease diagnosis results have higher influence weight, and the accuracy of phenotype prediction is improved.
Description
Technical Field
The invention relates to the technical field of neuro-image data analysis, in particular to a functional connection matrix processing system and device based on filtering method feature selection.
Background
With the development of science and technology, medical treatment, economy and the like, the living standard and the average life of people all over the world are improved. However, increasing competitive pressure, along with the tremendous mental stress on people, has led to an increase in the incidence of mental illness, year by year, and is one of the leading causes of death (see articles: van Waarde J A, scholte H S, van Oudheusden L J B, et al. A functional MRI marker major prediction of the outer of electrocomputes in a section and a project-resistant prediction [ J ]. Molecular therapy, 2015, 20 (5): 609-614.). How to find, diagnose and treat mental diseases more quickly and better has become a great concern for clinicians and researchers.
As a common neuroimaging technology, fMRI has been widely applied to the fields of clinical medicine, cognitive neuroscience, mental diseases and the like due to the characteristics of non-invasiveness, non-trauma, good space-time resolution, low cost and the like, and has now become an indispensable tool in the research of cognitive science, neuropsychiatry and neuroscience, and the understanding of people on complex pathogenesis and variable clinical biological differences of mental diseases is greatly deepened. Resting State Functional Magnetic Resonance Imaging (rs-fMRI) may reflect neural baseline activity of the brain when not tasked, and Functional Connectivity (FC) may be generated from rs-fMRI signals via some specific calculations. FC can effectively assess the degree of functional association between brain regions, is a common indicator in the field of brain imaging, and is often used as a feature for mental disease classification.
The general procedure for classifying mental disorders from rs-fMRI signals is: 1) Existing brain region templates, such as AAL (Automated chemical laboratory) were selected (see paper: tzourio-Mazoyer et al, "Automated atomic laboratory of activities in SPM using a macromolecular analytical partition of the MNI MRI single-subject bridge". NeuroImage.15 (1): 273-289.), yeo 2011 (see paper: yeo et al, the organization of The human center coded by interactive functional connectivity, J neurophysiol. 2011 Sep 106 (3): 1125-1165), and The like; 2) Extracting a mean time signal of each brain region to be tested based on the selected brain region template; 3) Calculating a functional connection matrix according to the average time signal of each brain region, for example, calculating a Pearson correlation coefficient (Pearson correlation coefficient) of rs-fMRI time signals of every two brain regions, and further calculating to obtain correlation Coefficient (COR) matrices of all brain regions; 4) Vectorizing the function matrix; 5) Selecting characteristics of the quantified result, and selecting partial characteristics with high correlation with the predicted phenotype; 6) And inputting the feature result after feature selection into a machine learning model or a deep learning model to predict diseases.
The feature selection is used as a critical step in the process, irrelevant redundant features can be removed, the number of features is reduced, an optimal feature subset is searched, the model is simplified, the running time of the model is reduced, and the accuracy of the model is improved. There are many mature methods for selecting characteristics, which can be classified into filtration, packaging and embedding. The basic idea is to calculate the correlation index quantity S of each feature and the class label for all S features i A 1, S i Sorting according to the sequence from large to small, setting a threshold value, and selecting the features with larger information quantity as the result of feature selection. For the classification problem, the main methods for measuring the correlation are F test, chi-square test and mutual information. In a conventional procedure for classifying mental diseases based on rs-fMRI signals, after feature selection is used, researchers directly input features larger than a certain threshold value into a machine learning model to perform disease phenotype prediction. However, inDuring feature selection, the correlation between each feature and the prediction phenotype is measured by different quantitative indexes, if only a threshold value is set for feature selection, the quantitative information of the correlation is not fully utilized, and the influence of each feature on final prediction is still determined by a machine learning model. According to the value of the quantitative index calculated by each feature, the influence of each feature on the final class prediction can be known to be different, and the feature with higher correlation with the class label has a larger weight influence on the final prediction result.
Therefore, a functional connection matrix processing system and device based on filtering method feature selection are provided.
Disclosure of Invention
In order to solve the above technical problems, the present invention provides a system and an apparatus for processing a function connection matrix based on a filtering method feature selection.
The technical scheme adopted by the invention is as follows:
a functional connectivity matrix processing system based on filter method feature selection, comprising:
the device comprises a test acquisition and preprocessing module, a test acquisition and preprocessing module and a test preprocessing module, wherein the test acquisition and preprocessing module is used for acquiring resting state brain function magnetic resonance images and disease diagnosis results after test preprocessing;
the brain region time sequence extraction module is used for extracting the time sequence of each brain region in each resting state brain function magnetic resonance image after being subjected to the pretreatment by using a brain image map;
the Pearson correlation coefficient calculation module is used for calculating the Pearson correlation coefficient of the time sequence of every two brain areas for each tested object to obtain a Pearson correlation coefficient matrix of each brain area;
the vectorization matrix module is used for vectorizing the pearson correlation coefficient matrix of each brain area for each tested object to obtain a vectorized pearson correlation coefficient matrix COR;
a quantitative correlation index calculation module for calculating quantitative correlation indexes S between each feature in all the vectorized Pearson correlation coefficient matrix COR and the disease diagnosis result by using a filtering method i And based on a predetermined thresholdSelecting a functional linkage feature COR having a high correlation with the disease diagnosis result sel And corresponding quantitative correlation index RELE sel ；
A feature conversion module for utilizing a corresponding quantitative relevance index RELE having a high relevance to the disease diagnosis result sel To the selected function connection feature COR sel Weighting to obtain a function connection matrix FC;
and the matrix prediction module is used for obtaining a prediction result through the function connection matrix FC.
Further, the preprocessing process of the resting state brain function magnetic resonance image in the subject acquisition and preprocessing module comprises: structural image decortication, cranial motion correction, temporal alignment, spatial smoothing, image registration, and/or spatial normalization.
Further, the brain image map in the brain region time sequence extraction module comprises a multi-modal brain map, a brain function map and/or a brain anatomy map.
Furthermore, the vectorization in the vectorization matrix module is to select a lower triangular element in the pearson correlation coefficient matrix, which does not include a diagonal, and flatten the lower triangular element into a one-dimensional vector.
Further, the calculation method of the filtering method in the quantitative correlation index calculation module is as follows:
combining N vectorized Pearson correlation coefficient matrixes COR, wherein the matrix dimension is N × S, S represents the feature dimension number after vectorization, and respectively calculating quantitative correlation indexes S between each feature and the disease diagnosis result i ；
A correlation index quantity S i Sorting according to the sequence from big to small, presetting a threshold value, and selecting S with high correlation with the disease diagnosis result sub Each feature is taken as a function connection feature and is marked as COR sel With matrix dimensions of N S sub ；
Recording the quantitative correlation index of the selected characteristics with high correlation with the disease diagnosis result, and recording as RELE sel With dimension 1 x S sub 。
Further, the weighting processing mode in the feature conversion module is as follows: FC = COR sel *δ*RELE sel Where δ is the scaling factor.
Further, the value of the scaling coefficient delta is 0.01 to 0.05.
Further, the matrix prediction module is specifically: and inputting the characteristic of the functional connection matrix FC into a machine learning model or a deep learning model to predict the phenotype.
The invention further provides a device for processing the function connection matrix based on the filtering method feature selection, which comprises a memory and one or more processors, wherein the memory stores executable codes, and the one or more processors are used for realizing the system for processing the function connection matrix based on the filtering method feature selection in any one of the embodiments when executing the executable codes.
The present invention also provides a computer-readable storage medium, which is characterized in that a program is stored thereon, and when the program is executed by a processor, the program implements a function connection matrix processing system based on filter method feature selection according to any one of the above embodiments.
The invention has the beneficial effects that: on the basis of feature selection by a filtering method, the selected features are further weighted by quantitative correlation indexes of the disease diagnosis results of the calculated features to better distinguish the influence effects of different features on the disease diagnosis results, so that the features with high correlation with the disease diagnosis results initially have higher influence weight, and the weighted features are put into a machine learning model for phenotype prediction to improve the accuracy of the phenotype prediction.
Drawings
FIG. 1 is a functional flow diagram of a functional connection matrix processing system based on filter method feature selection in accordance with the present invention;
FIG. 2 is a box plot of AUC values of the prediction results on the ABIDE dataset, obtained by randomly breaking up the ABIDE dataset into a test set and a training set, using different features for prediction, according to an embodiment;
fig. 3 is a block diagram of a function connection matrix processing apparatus based on filtering method feature selection according to the present invention.
Detailed Description
The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
Referring to fig. 1, a functional connection matrix processing system based on filtering method feature selection includes:
the device comprises a test acquisition and pretreatment module, a test acquisition and pretreatment module and a test analysis module, wherein the test acquisition and pretreatment module is used for acquiring resting state brain function magnetic resonance images after test pretreatment;
the preprocessing process of the resting state brain function magnetic resonance image in the tested acquisition and preprocessing module comprises the following steps: structural image skull removal, cranial movement correction, temporal alignment, spatial smoothing, image registration, and/or spatial normalization;
the brain region time sequence extraction module is used for extracting the time sequence of each brain region in each resting state brain function magnetic resonance image subjected to the pretreatment by using the brain image map;
the brain image map in the brain region time sequence extraction module comprises a multi-mode brain map, a brain function map and/or a brain anatomy map;
the Pearson correlation coefficient calculation module is used for calculating the Pearson correlation coefficient of the time sequence of every two brain areas for each tested object to obtain a Pearson correlation coefficient matrix of each brain area;
the vectorization matrix module is used for vectorizing the pearson correlation coefficient matrix of each brain area for each tested subject to obtain a vectorized pearson correlation coefficient matrix COR;
the vectorization mode in the vectorization matrix module is to select a lower triangular element which does not contain a diagonal line in a Pearson correlation coefficient matrix and flatten the lower triangular element into a one-dimensional vector;
quantitative correlation index calculationA module for calculating quantitative correlation index S between each feature in all the vectorized Pearson correlation coefficient matrix COR and the disease diagnosis result by using a filtering method i And selecting a functional connection characteristic COR with high correlation with the disease diagnosis result based on a preset threshold value sel And corresponding quantitative correlation index RELE sel ；
The calculation mode of the filtering method in the quantitative correlation index calculation module is as follows:
combining N vectorized Pearson correlation coefficient matrixes COR, wherein the matrix dimension is N x S, S represents the feature dimension degree after vectorization, and respectively calculating quantitative correlation indexes S between each feature and the disease diagnosis result i ；
A correlation index quantity S i Sorting according to the sequence from big to small, presetting a threshold value, and selecting S with high correlation with the disease diagnosis result sub The individual characteristics are taken as functional connection characteristics and are marked as COR sel With matrix dimension of N x S sub ；
Recording the quantitative correlation index of the selected characteristics with high correlation with the disease diagnosis result, and recording as RELE sel With dimension 1 × S sub 。
A feature conversion module for utilizing a corresponding quantitative relevance index RELE having a high relevance to the disease diagnosis result sel To the selected function connection feature COR sel Weighting to obtain a function connection matrix FC;
the weighting processing mode in the feature conversion module is as follows: FC = COR sel *δ*RELE sel Where δ is a scaling factor;
the value of the scaling coefficient delta is 0.01 to 0.05;
the matrix prediction module is used for obtaining a prediction result through the function connection matrix FC;
the matrix prediction module is specifically: and inputting the characteristic of the functional connection matrix FC into a machine learning model or a deep learning model to predict the phenotype.
The embodiment is as follows: a functional connectivity matrix processing system based on filtering feature selection, comprising:
the device comprises a test acquisition and preprocessing module, a test acquisition and preprocessing module and a Data processing module, wherein the test acquisition and preprocessing module is used for collecting preprocessed resting state Brain function magnetic resonance image Data from ABIDE (automatic Brain Imaging Data Exchange), and performing structural image skull removal, head movement correction, time alignment, space smoothing, image registration, space standardization and the like to obtain the resting state Brain function magnetic resonance image Data preprocessed by the test;
the ABIDE dataset contained 866 subjects whose 464 subjects were tested for normal and 402 subjects who were tested for autism.
The brain region time sequence extraction module is used for extracting the time sequence of each brain region in each resting state brain function magnetic resonance image after being subjected to the pretreatment by using a brain image map;
the brain image map in the brain region time sequence extraction module comprises a multi-mode brain map, a brain function map and/or a brain anatomy map;
AAL templates were selected for extraction of each brain region time series. The AAL template comprises 116 brain areas in total, and after the time sequence is extracted, a time sequence matrix of N116 is obtained, wherein N represents the time sequence length of the brain function magnetic resonance image in a resting state.
The Pearson correlation coefficient calculation module is used for calculating the Pearson correlation coefficient of the time sequence of every two brain areas for each tested object to obtain a Pearson correlation coefficient matrix of each brain area;
calculating a Pearson correlation coefficient between the time series of every two brain areas to obtain a Pearson correlation coefficient matrix of 116 × 116, wherein the Pearson correlation coefficient is calculated according to the following formula:
wherein r represents the Pearson correlation coefficient,
time i of time series X representing one of the brain regionsThe signal of the intermediate point is used as the signal,
the signal of the ith time point of the time series Y representing another brain region,
represents the average value of the time series X,
denotes the average value of the time series Y, i =1,2, \ 8230, and n, n denotes the number of time series signals.
The vectorization matrix module is used for vectorizing the pearson correlation coefficient matrix of each brain area for each tested subject to obtain a vectorized pearson correlation coefficient matrix COR;
the vectorization mode in the vectorization matrix module is to select a lower triangular element which does not contain a diagonal line in a Pearson correlation coefficient matrix and flatten the lower triangular element into a one-dimensional vector; all tested vectorized pearson correlation coefficient matrices have dimensions 866 × 6786.
A quantitative correlation index calculation module for calculating quantitative correlation indexes S between each feature in the vectorized Pearson correlation coefficient matrix COR and the disease diagnosis result by using a filtering method i And selecting a functional connection characteristic COR with high correlation with the disease diagnosis result based on a preset threshold value sel And corresponding quantitative correlation index RELE sel ；
The calculation mode of the filtering method in the quantitative correlation index calculation module is as follows:
combining N vectorized Pearson correlation coefficient matrixes COR, wherein the matrix dimension is N x S, S represents the feature dimension degree after vectorization, and respectively calculating quantitative correlation indexes S between each feature and the disease diagnosis result i ；
The correlation is assigned to the quantity S i Sorting according to the sequence from big to small, presetting a threshold value, and selecting S with high correlation with the disease diagnosis result sub A characteristic as a function ofConnect characteristics, mark as COR sel With matrix dimension of N x S sub ；
After the vectorization operation, feature selection is performed using a filtering method, preferably using an analysis of variance F test, and for each feature after vectorization, an analysis of variance F value is calculated based on the class label, the greater the F value, the more relevant the feature is to the disease diagnosis. Setting a threshold value, and selecting the characteristic with larger F value. Preferably, the 10% quantile of all the characteristic F values is used as a threshold value, and 10% of the characteristics are selected and recorded as COR sel Dimension 866 × 679, for subsequent calculations.
Recording the quantitative correlation index of the selected characteristics with high correlation with the disease diagnosis result, and recording as RELE sel With dimension 1 × S sub 。
A feature conversion module for utilizing a corresponding quantitative relevance index RELE having a high relevance to the disease diagnosis result sel To the selected function connection feature COR sel Weighting to obtain a function connection matrix FC;
connecting features COR for selected functions sel And weighting the corresponding analysis of variance F value representing each characteristic to obtain a functional connection matrix FC. The F-value may represent the correlation between each feature and the disease diagnosis, using RELE sel Expressed, its dimension is 1 × 679.
The weighting processing mode in the feature conversion module is as follows: FC = COR sel *δ*RELE sel Wherein δ is a scaling coefficient, and the value of the scaling coefficient δ is 0.01, so that the dimension of the obtained functional connection matrix FC is 866 × 679.
The matrix prediction module is used for obtaining a prediction result through the function connection matrix FC;
the matrix prediction module is specifically: and inputting the characteristic of the functional connection matrix FC into a machine learning model or a deep learning model to predict the phenotype.
For all the testees, connecting the selected functions with the characteristic COR sel Respectively performing the same operation as the calculated function connection matrix FC as a prediction characteristicThe following treatment is carried out: randomly disorganizing all the features according to a test, splitting the features into a test set (75%) and a training set (25%), using the training set for training a classifier, predicting on the test set after training is finished, and calculating an AUC (Area under the ROC Curve) value. Repeating the above processing steps 100 times, and respectively taking the COR with the function connection characteristics sel And drawing a box line graph with the AUC value obtained on the test set by taking the functional connection matrix FC as a characteristic, calculating a mean value and verifying the difference of the mean value by using T test, thereby verifying the improvement of the disease prediction effect of the functional connection matrix FC obtained by the method. Preferably, the classifier used is a linear SVC classifier with the main parameters set to: the regularization parameter is L1, the loss function is squared _ change, the penalty coefficient of the loss function is 1, the allowable deviation for stopping iteration is 0.0001, and the maximum iteration frequency is 1000;
COR with functional connection characteristics sel The boxplot of the distribution of AUC values on the prediction set with the functional connection matrix FC as features is shown in fig. 2, with the mean values as shown in the following table:
| COR sel | FC | |
| AUC mean | 0.659 | 0.709 |
COR with functional connection characteristics sel Performing T test calculation on AUC values obtained by performing 100 times of test results on the prediction set by taking the calculated functional connection matrix FC as a characteristic, wherein the calculated T value is11.440, P value 1.358e-23.
According to the boxplot, the mean value and the T test result, the FC obtained by the method provided by the invention is used as a characteristic to carry out phenotype prediction, so that the prediction effect can be effectively improved.
Corresponding to the embodiment of the functional connection matrix processing system based on the filtering method feature selection, the invention also provides an embodiment of a functional connection matrix processing device based on the filtering method feature selection.
Referring to fig. 3, an embodiment of the present invention provides a function connection matrix processing apparatus based on filter characteristic selection, including a memory and one or more processors, where the memory stores executable codes, and when the one or more processors execute the executable codes, the one or more processors are configured to implement a function connection matrix processing system based on filter characteristic selection in the foregoing embodiments.
The embodiment of the function connection matrix processing device selected based on the filtering method characteristics can be applied to any equipment with data processing capability, such as computers and other equipment or devices. The apparatus embodiments may be implemented by software, or by hardware, or by a combination of hardware and software. The software implementation is taken as an example, and as a logical device, the device is formed by reading corresponding computer program instructions in the nonvolatile memory into the memory for running through the processor of any device with data processing capability. In terms of hardware, as shown in fig. 3, a hardware structure diagram of an arbitrary device with data processing capability where a function connection matrix processing apparatus selected based on a filtering method feature is located according to the present invention is shown, where in addition to the processor, the memory, the network interface, and the nonvolatile memory shown in fig. 3, an arbitrary device with data processing capability where an apparatus is located in an embodiment may generally include other hardware according to an actual function of the arbitrary device with data processing capability, and details thereof are not described again.
The implementation process of the functions and actions of each unit in the above device is specifically described in the implementation process of the corresponding step in the above method, and is not described herein again.
For the device embodiment, since it basically corresponds to the method embodiment, reference may be made to the partial description of the method embodiment for relevant points. The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of the present invention. One of ordinary skill in the art can understand and implement it without inventive effort.
An embodiment of the present invention further provides a computer-readable storage medium, on which a program is stored, and when the program is executed by a processor, the system for processing a functional connection matrix based on filtering feature selection in the foregoing embodiments is implemented.
The computer readable storage medium may be an internal storage unit, such as a hard disk or a memory, of any data processing device described in any previous embodiment. The computer readable storage medium may also be any external storage device of a device with data processing capabilities, such as a plug-in hard disk, a Smart Media Card (SMC), an SD Card, a Flash memory Card (Flash Card), etc. provided on the device. Further, the computer readable storage medium may include both an internal storage unit and an external storage device of any data processing capable device. The computer-readable storage medium is used for storing the computer program and other programs and data required by the arbitrary data processing-capable device, and may also be used for temporarily storing data that has been output or is to be output.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A system for processing a function connection matrix based on filter method feature selection, comprising:
the device comprises a test acquisition and preprocessing module, a test acquisition and preprocessing module and a test preprocessing module, wherein the test acquisition and preprocessing module is used for acquiring resting state brain function magnetic resonance images and disease diagnosis results after test preprocessing;
the brain region time sequence extraction module is used for extracting the time sequence of each brain region in each resting state brain function magnetic resonance image subjected to the pretreatment by using the brain image map;
the Pearson correlation coefficient calculation module is used for calculating the Pearson correlation coefficient of the time sequence of every two brain areas for each tested object to obtain a Pearson correlation coefficient matrix of each brain area;
the vectorization matrix module is used for vectorizing the pearson correlation coefficient matrix of each brain area for each tested object to obtain a vectorized pearson correlation coefficient matrix COR;
a quantitative correlation index calculation module for calculating quantitative correlation indexes S between each feature in all the vectorized Pearson correlation coefficient matrix COR and the disease diagnosis result by using a filtering method i And selecting a functional connection characteristic COR with high correlation with the disease diagnosis result based on a preset threshold value sel And corresponding quantitative correlation index RELE sel ；
A feature conversion module for utilizing a corresponding quantitative relevance index RELE having a high relevance to the disease diagnosis result sel To the selected function connection feature COR sel Weighting to obtain a function connection matrix FC;
and the matrix prediction module is used for obtaining a prediction result through the function connection matrix FC.
2. The system as claimed in claim 1, wherein the preprocessing procedure of the brain function magnetic resonance image in resting state in the subject acquisition and preprocessing module includes: structural image decortication, cranial motion correction, temporal alignment, spatial smoothing, image registration, and/or spatial normalization.
3. The system according to claim 1, wherein the brain image map in the brain region time series extraction module comprises a multi-modal brain map, a brain function map, and/or a brain anatomy map.
4. The system of claim 1, wherein the vectorization in the vectorization matrix module is performed by selecting a lower triangular element of the pearson correlation coefficient matrix that does not include a diagonal and flattening the lower triangular element into a one-dimensional vector.
5. The system of claim 1, wherein the filtering method in the quantitative correlation index calculation module is calculated by:
combining N vectorized Pearson correlation coefficient matrixes COR, wherein the matrix dimension is N × S, S represents the feature dimension number after vectorization, and respectively calculating quantitative correlation indexes S between each feature and the disease diagnosis result i ；
The correlation is assigned to the quantity S i Sorting according to the sequence from big to small, presetting a threshold value, and selecting S with high correlation with the disease diagnosis result sub The individual characteristics are taken as functional connection characteristics and are marked as COR sel With matrix dimensions of N S sub ；
Recording the quantitative relevance index of the selected characteristics with high relevance to the disease diagnosis result, and recording as RELE sel With dimension 1 × S sub 。
6. The system of claim 1, wherein the feature transformation comprises a function connection matrix, and wherein the function connection matrix comprises a function connection matrix of a plurality of function connection matricesThe weighting processing mode in the module is as follows: FC = COR sel *δ*RELE sel Where δ is the scaling factor.
7. The system according to claim 6, wherein the scaling factor δ is 0.01 to 0.05.
8. The system of claim 1, wherein the matrix prediction module is specifically configured to: and inputting the characteristic of the functional connection matrix FC into a machine learning model or a deep learning model to predict the phenotype.
9. A filter characteristic selection-based function connection matrix processing apparatus, comprising a memory and one or more processors, wherein the memory stores executable code, and the one or more processors, when executing the executable code, implement a filter characteristic selection-based function connection matrix processing system according to any one of claims 1 to 8.
10. A computer-readable storage medium, having stored thereon a program which, when executed by a processor, implements a filter feature selection based functional connection matrix processing system according to any one of claims 1 to 8.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211070002.9A CN115147417B (en) | 2022-09-02 | 2022-09-02 | Function connection matrix processing system and device based on filtering method feature selection |
| US18/360,796 US11989883B2 (en) | 2022-09-02 | 2023-07-27 | Functional connectivity matrix processing system and device based on feature selection using filtering method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211070002.9A CN115147417B (en) | 2022-09-02 | 2022-09-02 | Function connection matrix processing system and device based on filtering method feature selection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115147417A CN115147417A (en) | 2022-10-04 |
| CN115147417B true CN115147417B (en) | 2022-11-15 |
Family
ID=83416127
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211070002.9A Active CN115147417B (en) | 2022-09-02 | 2022-09-02 | Function connection matrix processing system and device based on filtering method feature selection |
Country Status (2)
| Country | Link |
|---|---|
| US (1) | US11989883B2 (en) |
| CN (1) | CN115147417B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118247682B (en) * | 2024-05-24 | 2024-08-16 | 武汉大学 | Land coverage classification method and system based on first satellite of stars |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108257657A (en) * | 2016-12-28 | 2018-07-06 | 复旦大学附属华山医院 | The data analysing method of magnetic resonance detection based on the prediction of disturbance of consciousness patient consciousness recovery |
| WO2019069955A1 (en) * | 2017-10-03 | 2019-04-11 | 株式会社国際電気通信基礎技術研究所 | Differentiation device, differentiation method for depression symptoms, determination method for level of depression symptoms, stratification method for depression patients, determination method for effects of treatment of depression symptoms, and brain activity training device |
| CN109886498A (en) * | 2019-03-01 | 2019-06-14 | 北京邮电大学 | A kind of EMD-GRU Short-Term Load Forecasting Method based on feature selecting |
| CN114334140A (en) * | 2022-03-08 | 2022-04-12 | 之江实验室 | Disease prediction system and device based on multi-relation function connection matrix |
| CN114816321A (en) * | 2022-04-22 | 2022-07-29 | 吉林大学 | Unsupervised method for determining importance of user perception index |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6722845B2 (en) * | 2017-10-03 | 2020-07-15 | 株式会社国際電気通信基礎技術研究所 | Discriminating apparatus, depressive symptom discriminating method, depressive symptom level determining method, depressive patient stratification method, depressive symptom therapeutic effect determining method, and brain activity training apparatus |
| US20210401289A1 (en) * | 2018-10-11 | 2021-12-30 | Advanced Telecommunications Research Institute International | Brain functional connectivity correlation value adjustment method, brain functional connectivity correlation value adjustment system, brain activity classifier harmonization method, brain activity classifier harmonization system, and brain activity biomarker system |
| EP3977463A1 (en) * | 2019-06-03 | 2022-04-06 | Biomax Informatics AG | Computing a pathological condition |
| WO2021205996A1 (en) * | 2020-04-06 | 2021-10-14 | 株式会社国際電気通信基礎技術研究所 | Brain functional connectivity correlation value clustering device, brain functional connectivity correlation value clustering system, brain functional connectivity correlation value clustering method, brain functional connectivity correlation value classifier program, brain activity marker classification system and clustering classifier model for brain functional connectivity correlation values |
| EP4157080A4 (en) * | 2020-05-29 | 2024-06-19 | The Board of Regents of the University of Texas System | Network-based functional imaging output for evaluating multiple sclerosis |
-
2022
- 2022-09-02 CN CN202211070002.9A patent/CN115147417B/en active Active
-
2023
- 2023-07-27 US US18/360,796 patent/US11989883B2/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108257657A (en) * | 2016-12-28 | 2018-07-06 | 复旦大学附属华山医院 | The data analysing method of magnetic resonance detection based on the prediction of disturbance of consciousness patient consciousness recovery |
| WO2019069955A1 (en) * | 2017-10-03 | 2019-04-11 | 株式会社国際電気通信基礎技術研究所 | Differentiation device, differentiation method for depression symptoms, determination method for level of depression symptoms, stratification method for depression patients, determination method for effects of treatment of depression symptoms, and brain activity training device |
| CN109886498A (en) * | 2019-03-01 | 2019-06-14 | 北京邮电大学 | A kind of EMD-GRU Short-Term Load Forecasting Method based on feature selecting |
| CN114334140A (en) * | 2022-03-08 | 2022-04-12 | 之江实验室 | Disease prediction system and device based on multi-relation function connection matrix |
| CN114816321A (en) * | 2022-04-22 | 2022-07-29 | 吉林大学 | Unsupervised method for determining importance of user perception index |
Non-Patent Citations (4)
| Title |
|---|
| Abnormal Local Brain Activity Beyond the Pain Matrix in Postherpetic Neuralgia Patients: A Resting-State Functional MRI Study;Song Cao等;《Pain Physician》;20170228;第E303-E314页 * |
| Attentional Connectivity-based Prediction of Autism Using Heterogeneous rs-fMRI Data from CC200 Atlas;Yaya Liu等;《Experimental Neurobiology》;20200128;第27-37页 * |
| 基于高阶最小生成树的脑网络分析及对阿兹海默氏症患者的分类;郭浩等;《计算机应用》;20171110(第11期);第3339-3344页 * |
| 阿尔茨海默病患者脑功能连接变化分析及分类研究;杜宇航;《中国优秀硕士学位论文全文数据库医药卫生科技辑》;20200815;E071-10 * |
Also Published As
| Publication number | Publication date |
|---|---|
| US20240078678A1 (en) | 2024-03-07 |
| US11989883B2 (en) | 2024-05-21 |
| CN115147417A (en) | 2022-10-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Janghel et al. | Deep convolution neural network based system for early diagnosis of Alzheimer's disease | |
| Liu et al. | Towards clinical diagnosis: Automated stroke lesion segmentation on multi-spectral MR image using convolutional neural network | |
| CN108288070B (en) | Neural fingerprint extraction and classification method and system | |
| Gull et al. | Artificial intelligence in brain tumor detection through MRI scans: advancements and challenges | |
| Pan et al. | Characterization multimodal connectivity of brain network by hypergraph GAN for Alzheimer’s disease analysis | |
| Sato et al. | Identification of psychopathic individuals using pattern classification of MRI images | |
| Sun et al. | Dual-functional neural network for bilateral hippocampi segmentation and diagnosis of Alzheimer’s disease | |
| Al-Adhaileh | Diagnosis and classification of Alzheimer's disease by using a convolution neural network algorithm | |
| Zhao et al. | Feature fusion via hierarchical supervised local CCA for diagnosis of autism spectrum disorder | |
| CN103942567A (en) | Automatic discrimination analysis method of mild cognitive impairment based on support vector machine | |
| Collazos-Huertas et al. | Instance-based representation using multiple kernel learning for predicting conversion to Alzheimer disease | |
| CN115147417B (en) | Function connection matrix processing system and device based on filtering method feature selection | |
| Yang et al. | Diagnosis of Parkinson’s disease based on 3D ResNet: The frontal lobe is crucial | |
| Sarkar et al. | Computational Intelligence Approach to improve the Classification Accuracy of Brain Tumour Detection | |
| Janghel | Deep-learning-based classification and diagnosis of Alzheimer's disease | |
| CN116759096B (en) | Disease prediction system based on multiple comparison function connection matrix | |
| CN114373095A (en) | Alzheimer disease classification system and method based on image information | |
| Davuluri et al. | A survey of different machine learning models for Alzheimer disease prediction | |
| Erdaş et al. | A deep learning method to detect Parkinson’s disease from MRI slices | |
| Stasiak et al. | Application of convolutional neural networks with anatomical knowledge for brain MRI analysis in MS patients | |
| CN112561935A (en) | Method, device and equipment for identifying Alzheimer's disease | |
| Matlani | BiLSTM-ANN: early diagnosis of Alzheimer’s disease using hybrid deep learning algorithms | |
| CN111938644A (en) | Early diagnosis method, device and system for Alzheimer disease based on LLE | |
| CN111951228B (en) | Epileptogenic focus positioning system integrating gradient activation mapping and deep learning model | |
| Padole et al. | Early Detection of Alzheimer’s Disease Using Graph Signal Processing and Deep Learning. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |